Are all your recent patches in CVS?
Also, would you check with Georg to see whether
your latest patches fix the DABUSB problems
he's reported? If the DABUSB problem that apparantly
surfaces as a result of using mixed isochronous and
bulk transfers is fixed, we may be ready to evaluate
whether it's time to standardize on one UHCI HCD
(I know that Georg is working hard to get usb-uhci
cleared of remaining bugs, as well).
Lastly, perhaps now would be a good time to request
that everyone who has experienced bugs with uhci try
to reproduce them with your latest code. It'd be
great if we can get a comprehensive uhci bug list
together. For the 2.4 launch, I recall that Randy
sifted through the e-mail archives for bug reports
and contact the bug reporters. Perhaps we should
do this now for uhci bugs?
Miles
Johannes Erdfelt wrote:
>
> I have a new version of my uhci.c patch. Here's the list of changes:
>
> - Fixed bulk urb queuing!
> - Fix a bug in lprintk (no newline)
> - Remove some debugging code (urbp->elements, root_hub)
>
> With the default MTU, ttcp and the bulk queuing enabled version of usbnet:
>
> ttcp-t: 16777216 bytes in 29.19 real seconds = 561.24 KB/sec +++
> ttcp-t: 2048 I/O calls, msec/call = 14.60, calls/sec = 70.15
> ttcp-t: 0.0user 0.1sys 0:29real 0% 0i+0d 0maxrss 0+2pf 0+0csw
>
> With a large MTU (8k) and the same software:
>
> ttcp-t: 16777216 bytes in 20.69 real seconds = 792.06 KB/sec +++
> ttcp-t: 2048 I/O calls, msec/call = 10.34, calls/sec = 99.01
> ttcp-t: 0.0user 0.0sys 0:20real 0% 0i+0d 0maxrss 0+2pf 0+0csw
>
> This is also removing the skb_reserve call in usbnet.c. David, is that
> correct?
>
> I'm still continuing testing.
>
> JE
>
> ------------------------------------------------------------------------
>
> uhci-20010328.patchName: uhci-20010328.patch
> Type: Plain Text (text/plain)
_______________________________________________
[EMAIL PROTECTED]
To unsubscribe, use the last form field at:
http://lists.sourceforge.net/lists/listinfo/linux-usb-devel
